INTERNATIONAL FOOD POLICY RESEARCH INSTITUTE

Natural Disasters and their Labor Market Consequences: Evidence from the 1998 Flood in Bangladesh

Valerie Mueller (IFPRI) Agnes Quisumbing (IFPRI)

WEAI Conference, Vancouver June 30, 2009

Background

  Natural disasters have long-term consequences on agricultural growth (Sachs, 2001; Gallup and Sachs, 2000)

  Natural disasters impede accumulation of physical and capital stock (Yamauchi et al., 2008a, 2008b; Skoufias, 2003)

  Climate change increasing droughts, floods, and sea level (IPCC, 2007)

  Large development literature on constraints of households to cope with shocks

Research Objective

  Few studies focus on role of migration and labor markets in coping with natural disasters

  Migration and employment in RNF sector used to cope with shocks

  We evaluate how resilient labor markets in rural Bangladesh were to the 1998 “flood of the century” (short-term vs. long-term)

  We also look at factors that mitigated the damages

  Panel household survey collected for flood impact assessment (del Ninno et al., 2001)

Theoretical Insight on Disasters and Markets

  Consumption smoothing literature shows failure of the PIH model in developing countries

  Literature lacks insight on long-term impacts of widespread shocks •  Community risk-sharing less likely as informal

creditors overburdened (Townsend, 1994) •  Underinvestment because of risk aversion and

asset depletion (Rosenzweig and Binswanger, 1993)

Long-term Disaster Impacts on Labor Markets

  Complementarity of capital and labor crucial   Decline in agricultural labor demand (most

focus on this effect)   Non-farm labor market also vulnerable if

migration is costly or local labor surplus from shock

The 1998 “Flood of the Century”

  Bangladesh use to annual floods   1998 flood most severe due to duration and

coverage   Food assistance programs available   Short-term effects on consumption, nutrition,

assets, debt (del Ninno et al., 2001)   Recent studies focus on long-term impacts on

consumption, and physical and human capital accumulation (Yamauchi et al., 2008a, 2008b; Quisumbing, 2005a, 2005b)

Data

  757 households in 126 villages from November 1998-May 2004 (four rounds)

  Focus on casual labor market module   Use previous month wage data*   Plot level information on normal and realized

flood depth (feet)   Village flood measure averages the deviation

of the 1998 depth from the normal   Mitigation: Irrigation, soil type, credit,

distance to nearest market

Empirical Strategy

  Daily wage regression   Covariates

•  Individual labor supply characteristics •  Flood measure •  Control for initial labor market conditions in

1997 •  Include thana, month, and year fixed effects

  Pooled OLS and Random effects models*   Short-term vs. Long-term models   Thana clustering for arbitrary correlation of

flood impacts

Baseline results

Pooled RE Pooled RE

OLS GLS OLS GLS

1998 Flood shock -0.019* -0.017* -0.009 -0.007

(0.009) (0.009) (0.010) (0.012)

1998 Flood shock*Year 1999 dummy -0.021 -0.020

(0.018) (0.019)

1998 Flood shock*Year 2004 dummy -0.047* -0.044*

(0.022) (0.026)

Observations 1470 1470 1470 1470

R-squared 0.30 0.30 0.30 0.30

Agricultural vs. Non-Agricultural Wages

Pooled RE Pooled RE

OLS GLS OLS GLS

1998 Flood shock -0.009 -0.009 0.014 0.014

(0.013) (0.013) (0.014) (0.014)

1998 Flood shock*Year 1999 dummy -0.036 -0.036

(0.023) (0.025)

1998 Flood shock*Year 2004 dummy -0.069** -0.065***

(0.023) (0.025)

1998 Flood shock*Agriculture -0.015 -0.013 -0.038** -0.036***

(0.018) (0.018) (0.014) (0.015)

1998 Flood shock*Year 1999 dummy* 0.027 0.027

Agriculture (0.017) (0.017)

1998 Flood shock*Year 2004 dummy 0.034*** 0.032***

Agriculture (0.006) (0.007)

Observations 1470 1470 1470 1470

R-squared 0.30 0.30 0.31 0.31

Mitigation

  Percent Irrigation: Ability to shift cultivation to the dry season

  Drainage capacity: soil type can reduce the scope of the crop loss

  Presence and scale of informal credit system: Can reduce the distress sale of assets

  Proximity to markets and bazaars: provide workers access to additional outlets for employment

Findings on Mitigation

  Areas with clay soil were most severely affected in the short term

  Irrigation and credit access might mitigate flood impacts •  Irrigation variable didn’t vary over time, so

couldn’t identify statistically significant effect •  Credit access results weren’t robust

  Labor markets closer to the weekly market or bazaar were less affected than those further away in the long-term

Conclusion

  Severe flood caused 4-5% decline in real wages

  Emergency relief programs might have protected individuals in the short-term but not the long-term

  Non-agricultural markets suffered perhaps due to their dependence on the recovery of other markets

  Migration could possibly mitigate these impacts suggesting policies aimed at reducing moving and search costs may be a temporary solution to recover from major flood